ARTICLE IN PRESS

+ MODEL

Journal of Archaeological Science xx (2006) 1e18

http://www.elsevier.com/locate/jas

Blind tests shed light on possibilities and limitations for identifying

stone tool prehension and hafting
V. Rots a,*, L. Pirnay a, Ph. Pirson a, O. Baudoux b
a
Prehistoric Archaeology Unit, Katholieke Universiteit Leuven, Geo-Institute, Celestijnenlaan 200E, 3001 Heverlee, Belgium
b
Chercheurs de la Wallonie, Prehistosite de Ramioul, Rue de la Grotte 128, 4400 Flemalle, Belgium
Received 10 January 2005; received in revised form 2 May 2005; accepted 27 October 2005

Abstract

Prehensile wear has never formed the focus of a blind test in microwear studies and doubts remain about the formation, identification and
interpretation of diagnostic prehension and hafting wear. The results of the presented blind tests demonstrate that prehension and hafting traces
do form and that their formation is sufficiently systematic and patterned to allow valid and reliable interpretations. A combined approach, in-
volving macroscopic, low power and high power analyses, is suggested as the most meaningful approach for consistent inferences.
2005 Elsevier Ltd. All rights reserved.

Keywords: Blind test; Microwear; Hafting; Prehension; Stone tool

1. Context importance (e.g. [7]), prehensile wear was not expected to

People have always been interested in what stone tools were
used for. Semenov [25] was the first to systematically deal with
the question and to come up with a microscopic technique that
made answers conceivable. Since then, use-wear analysis has
gone a long way. Different levels of magnification were tested
(e.g. [13,5,6,1,10,3]) and a methodology gradually developed.
Blind tests were crucial throughout this development. A blind
test is an objective means to evaluate the accuracy of informa-
tion retrieved by a specific method. Several blind tests were
executed at the beginning of functional research, a few apply-
ing low magnification [15,26,27], several applying high mag-
nification (e.g. [8,4,9,11,29,2]). Recently, a blind test was
also undertaken for residue analysis [31].
Prehensile wear, which is defined as traces resulting from
either prehension (hand-held use without any intermediate ma-
terial) or hafting (the use of a wrapping of some sort or the at-
tachment of a handle to the lithic tool) has never before
formed the focus of a blind test. Despite an agreement on its
* Corresponding author. Tel.: 32 16 326435; fax: 32 16 326400.
E-mail address: veerle.rots@geo.kuleuven.be (V. Rots).
form, at least not in a consistent and systematic way. Only
Odell attempted to propose valid distinctive criteria (e.g.
[12,14,15]). Recently, the work of Odell was extended on
a larger and more systematic basis [16]. Based on a large
experimental reference set, criteria were proposed for a distinc-
tion between hand-held and hafted tools and for the interpre-
tation of the hafting arrangement used [16e18,20]. Several
blind tests were undertaken throughout this research and
they form an important argument in the discussion concerning
the formation and interpretability of prehensile wear. Blind
tests are also extremely instructive and provocative for the an-
alyst and they can pinpoint interpretative problems not real-
ized before. Blind tests thus form an important tool for
quality improvement and accuracy of the designed method.
Therefore, tests should be undertaken on a regular basis,
from the start of ones research, despite the risk of bad test re-
sults. However, it is evident that in order to make a strong
case, results need to be good. This brings one to an impasse.
On the one hand, experimental results need to be examined
on their validity as quickly as possible, but on the other
hand, test results may suffer from inexperience and incomplete
methodological groundwork. Based on these considerations, it

0305-4403/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2005.10.018
 ARTICLE IN PRESS
+ MODEL

2 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Notes for the understanding of the variables included

in the tables

Haft limit:
value concerns the distance from the butt to the
haft limit and represents the intrusion of the
stone tool in its haft (if it was hafted).
Haft material:
material out of which the handle is made in con-
trast to other materials that may be used for fix-
ation, like bindings or resin.
Wrapping:
material a tool may be wrapped in after which it
may be mounted on a handle.
Haft type:
two main haft types are relevant, a juxtaposed
one in which the stone tool is mounted next to
the handle and a male one in which the stone
tool is inserted in a handle. The latter can be
further divided into a male handle sensu strictu
when the tool is inserted into a hole or a male
split haft when the tool is inserted into a cleft.
Hafting method:
describes the contact between the lithic tool and
its haft, when it is direct, there is no material in
between the stone tool and its haft, when it is in-
direct, there is (e.g. wrapping).
Tool placement: Plate 1. Examples of hafted stone tools: indirect hafting in male antler haft
position of the tool with regard to the handle: with leather wrapping of BT19, direct hafting in male split antler haft of
latero-distal at the end of a bent handle, termi- BT20 and fixation with leather bindings.
nal at the end of a straight handle, lateral at
the side of the handle. analyst re-cleaned all tools before starting the analysis. For
Tool direction: cleaning, a short immersion in an HCl-solution (10%) was
orientation of the stone tool with regard to the used and during analysis tools were cleaned with acetone or
axis of the handle: transversal, axial or oblique. alcohol. For the first test, a few strict guidelines were formu-
Orientation of the active part: lated: if the tools were used, it could be for one function
orientation of the working edge with regard to only, tools were to be used e in the hand or hafted e with
the axis of the handle: perpendicular, parallel a minimal duration of 30 min, tools needed to be freshly pre-
or oblique to it. pared, without possible external friction (they could not be
transported, trampled, etc.). All other parameters were left

was decided to undertake blind tests at different stages during

the research, from the beginning onwards. Interpretative prob-
lems could thus be highlighted, making a direct contribution to
a further methodological development. Mistakes should be
viewed in this light.

2. Methods

Three separate and consecutive blind tests were undertaken.

L. Pirnay, P. Pirson and O. Baudoux produced, hafted (Plate 1)
and used (Plate 2) each stone tool without providing any infor-
mation to the functional analyst (V. Rots). Only flint was used
as a raw material (both fine- and coarse-grained). Tools were
handed over to the analyst after de-hafting and cleaning. The Plate 2. Sawing dry antler with BT20.
 ARTICLE IN PRESS
+ MODEL
V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
up to the experimenters (i.e. worked material, activity, hafting,
etc.). For the remaining two tests, no restrictions were formu-
lated. While the first test is an explorative test, focused on the
general interpretability of hafting wear, both other tests may
be considered as testing the method itself for deriving inferen-
ces concerning prehension and hafting. For none of the tests,
guidelines were provided for the minimal number of hand-
held or hafted tools to be included, this choice was left up
to the experimenters. The analyst was thus unaware of whether
hand-held or hafted tools were included at all.
All interpretations are based on comparisons with an exten-
sive experimental reference set (about 400 experimental tools)
that is considered to be representative for the variability of pre-
hension and hafting traces [16,20]. The experiments were per-
formed outside, in non-laboratory conditions. The impact of
several variables on the formation of wear traces was exam-
ined [16] and a distinction was made between dominant and
secondary variables, based on their importance for the hafting
trace formation process. Dominant variables include the haft-
ing arrangement, hafting material, use motion and worked ma-
terial [23], and the secondary variables include raw material
coarseness, tool morphology and others. The hafting arrange-
ment includes haft type (male: insertion in a handle, male
split: insertion into a cleft, juxtaposed: fixed next to a handle),
hafting method (direct or indirect (e.g. due to a wrapping) con-
tact with the haft), tool placement (terminal: at the end of the
handle, lateral: at the side of the handle, latero-distal: at the
end of a bent handle); tool direction (in the axis of the handle
or not), orientation of the active part (working edge) with re-
gard to the axis of the handle (parallel, perpendicular or obli-
que) (examples in Plate 1).
Traces were inventoried based on a detailed registration
system that considered 21 units, of which 11 are located on
the dorsal face, nine on the ventral face and one on the butt
(Plate 3, [16]). The units are dispersed over three tool parts,
the proximal, medial and distal tool portions. The medial
area corresponds to the area around the haft limit, which is
a necessary division given the distinct nature of the wear traces
Plate 3. Schematic representation of a stone tool with indications of the divisions used for wear registration.
3
in this zone. Further divisions concern the distinction between
the edges, ridges and surfaces.
3. Equipment
For low power analysis, a stereoscopic microscope (Wild,
magnifications 6
e100
) was used according to the princi-
ples set out by Tringham et al. [28] and further elaborated
by Odell [13]. We focused on scarring mainly, but other kinds
of wear like polish, rounding and striations were examined as
well. For high power analysis, tools were analyzed with a met-
allurgical microscope (Olympus, magnifications 50
e500
),
using bright field illumination, according to Keeley [6]. Polish,
rounding, striations and scarring were examined.
4. Distinctive criteria for distinguishing prehension
and hafting
The most important distinctive criteria for identifying
prehension and hafting are summarized, for a more elaborate
discussion we refer to Rots [16e20]. A distinction of hafted
tools from hand-held tools is possible based on the presence
of a clear limit between the used and hafted tool portion in
the form of a sudden start of (different) wear traces, like the
abrupt start of scars or polish, the presence of a bright spot
(localized well-developed polish spots) concentration prefera-
bly in association with scarring (Plate 4, [17]), etc. In the case
of resin-hafted tools, the abrupt limit generally consists of an
absence of traces in the hafted area with the exception of pos-
sible resin friction bright spots [17]. The latter are mainly
formed during tool extraction. Hand-held tools do not show
an abrupt interruption of the wear pattern and polish domi-
nates, the morphology of which is determined by the worked
material given that the particles that detach from the worked
material during use cause the prehension polish formation.
The extent and intensity of a prehension polish is determined
by the amount of particles that detach during use, in the case
of dirty materials such as bone/antler or schist, the
 ARTICLE IN PRESS
+ MODEL
4 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
Plate 4. Hafting bright spot associated with scarring on an experimental tool
used for adzing wood (200
).
prehension polish can be very extensive (Plate 5). For hafted
tools, polish and scarring are the dominant wear traces, includ-
ing bright spots [17], while striations and rounding are infre-
quent. The polish tends to follow the microtopography and
can be interpreted based on similar traits as those used for
use-polish interpretations. Also the scar characteristics may
be typical for a certain hafting arrangement, in particular bind-
ings result in distinctive sliced (or sliced into scalar) scars with
a curved initiation (Plate 6).
5. Distinction of prehensile wear from production and
use-wear
Production wear is easy to distinguish from prehensile
wear, as it is always associated with a technological feature,
Plate 5. Extensive prehension polish formation on the ventral medial edge as
a result of a contact with the hand during experimental antler working (200
).
Plate 6. Experimental sliced into scalar scar (curved initiation) resulting from
hafting, typically formed on edges that are in contact with bindings (50
).
like the butt, bulbar scar or retouch. It consists of minor polish
formation, scarring and striations (Plate 7). The wear charac-
teristics are indicative for the contact material, be it the ham-
mer that is used or the stone material itself in the case of
friction with the core upon detach (e.g. on the bulb or ventral
butt) [24].
Use-wear can be distinguished based on the clear impact of
the wear traces on the edge, the directional aspect of the traces
and the occurring types of associations between wear traces
[19]. An association of polish and rounding is for instance fre-
quent (Plate 8), while this is not the case for hafting wear.
Striations are also far more important in the case of use-
wear than in the case of prehensile wear. In addition, bright
spots rarely form as a result of use, but they may form when
a flint particle detaches from the used edge and causes a short
though intense friction with the used edge. These use-related
Plate 7. Knapping striations from the friction with a stone hammer on the im-
pact point of the butt of an experimental tool (200
).
 ARTICLE IN PRESS
+ MODEL
V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
Plate 8. Clear rounding associated with use-wear polish on the scraper-head of
an experimental tool used to scrape tanned sheep hide (1 h, 200
).
bright spots are always integrated within the use-wear traces
(Plate 9, [17]).
6. Results
The results of each blind test are presented separately given
the specific conditions under which each one took place. All
errors are discussed systematically according to their rele-
vance and importance for the method that is being designed.
The results are summarized in two series of tables. A first se-
ries of tables (Tables 1e3) concerns a summary of the exper-
imental data and the interpretation that was obtained for each
aspect through a functional analysis. All variables that were
taken into account during the experiment, such as the exact
tool use and the different aspects of the hafting arrangement
Plate 9. Bright spot (from the friction with a flint particle) integrated in a hide
use-wear polish on the ventral scraper-head of an experimental tool (200
).
5
(see also notes for tables), if relevant, are dealt with separately.
The second series of tables (Tables 4e8) concerns the evalua-
tion of the accurateness of the interpretation for each tool and
each variable. In the discussion, suggestions are proposed for
improving interpretations. Focus is on the distinction between
hand-held and hafted tools on the one hand, and the adequate
interpretation of the hafting material and arrangement on the
other hand. Tool use is considered to be an important assess-
ment only when it has consequences for correct interpretations
of prehension, hafting or the hafting material and arrangement
used.
For certain aspects, an error range is allowed. Exact use du-
rations, for instance, are not considered to be important and
only relative assessments are made (short: <10 min, moderate:
10e30 min, long: 30e60 min., very long: 1 h). For the local-
ization of haft limits, errors of up to 0.5 cm are accepted. For
each interpretation, its degree of certainty was recorded sys-
tematically on a scale of five during the analysis (0: uncertain
to 4: certain). These certainty levels are only referred to when
relevant.
6.1. Blind test 1
The experimental data and the respective wear interpreta-
tion per tool following the functional analysis are included
in Table 1. Scores are summarized in Table 4.
On a total of eight tools, three were interpreted correct on
all levels, for three others minor mistakes were made, while
two last ones were misinterpreted (Table 4). These results
are not fully satisfactory yet, but they indicate that hafting
traces are produced and that they are interpretable. The suc-
cess rate is too high to be coincidental and the kind of infor-
mation retrieved is too detailed to be a result of guessing.
6.1.1. Error 1: prehension versus hafting
The two most important interpretative mistakes concern the
misinterpretation of the prehensile mode of two tools, BT1 and
BT5. Both interpretations were marked as uncertain, but a dis-
cussion is still relevant. In both cases the error is a consequence
of the worked material, schist. Schist is one of the few materials
that may cause a macroscopically visible polishing ([16,24]),
which is potentially confused with a hafting against wood. In
juxtaposed arrangements, such a polishing occurs on one sur-
face only (i.e. the face in contact with the haft), but in male ar-
rangements, it may occur on both surfaces (i.e. both faces are in
contact with the haft). Given the assumed worked material, the
observed macroscopic polishing was understood as a conse-
quence of use instead of hafting. The error was thus not caused
by the hafting arrangement despite it being equal for both tools
(i.e. male wooden haft). This is supported by the correct inter-
pretation of BT7, another male-hafted tool. The latter was
hafted in antler instead of wood and antler hafts do not cause
a macroscopically visible polish, which allows for the hafting
arrangement to be correctly inferred.
The correct identification of a limit between the active and
non-active (i.e. hafted) part on both tools demonstrates that
this mistake can potentially be avoided. Experiments
 6
Table 1
Blind test 1: experimental details and proposed interpretation following a functional analysis of the wear traces ( included experimental data are correctly inferred based on wear analysis,  no data are
provided by the functional analyst, X irrelevant category for the tool in question)
Results BT1 BT2 BT3 BT4 BT5 BT6 BT7 BT8
Tool type Drill Drill Endscraper Burin Endscraper Burin Endscraper Burin
Interpretat Exp Interpr Interpret Interpreta Interpre Interpre Interpreta
Exp details Exp details Exp details Exp details Exp details Exp details Exp details
ion details etation Interpretation ation tion tation tation tion
Used part distal part + distal part + distal part + distal part + distal part + distal part + distal part + distal part +
U Worked dry tanned wood (or hide bone/ dry yew hide on dry deer bone/
dry antler schist schist + antler schist + + +
s material leather + ochre) antler wood wood bone antler
e

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Action drilling grooving drilling + scraping + grooving + scraping + grooving + scraping + grooving +
Use duration 30min + 30min + 30min + 35min + 1h + 1h 35min + 1h 15min + 1h 30min moderate
Hand-held /
hafted hand-held hafted + hafted + hafted + hafted hand-held hafted + hafted + hafted +
hafted
Hafted /

ARTICLE IN PRESS
proximal proximal proximal proximal proximal proximal proximal
hand-held + + proximal part + + + + +
part part part part part part part
part
P
7&
r Haft limit 26mm 22mm 30mm 35mm 36mm 39mm 8 & 71mm 37mm 37mm 38mm 58mm 33mm 52mm 31mm 42mm
67mm
e
Haft leather
h wood wood + wood + + wood wood antler antler wood wood bone
material bindings

+
e
Contact zone

MODEL
n both ventral + ventral + both + both dorsal + both + ventral dorsal
haft
s
Wrapping X X + X + X + X X leather X + leather
i
o Contact zone
X X + X + X + X X ventral X + both
n wrapping
Bindings leather leather + leather + leather + leather leather vegetal leather + leather +
/
Contact zone
edges dorsal + dorsal + both + edges ventral ventral edges + dorsal ventral
h bindings
a
f Fixation X X + X + X + X X + X + X +
t Contact zone
i X X + X + X + X X + X + X +
fixation
n
g Haft type male split juxtaposed + juxtaposed + male + male split juxtaposed + male split + juxtaposed +
Hafting
direct direct + direct + direct + direct direct + direct + indirect direct
method
Tool
terminal terminal + terminal + terminal + terminal terminal + terminal + terminal +
placement
Tool
axial axial + axial + axial + axial axial + axial + axial +
direction
Orientation perpendicu perpendicu
parallel parallel + perpendicular + parallel + parallel + + parallel +
active part lar lar
 ARTICLE IN PRESS
+ MODEL
V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
systematically indicate that no limits are identifiable in the
case of a hand-held use, while it forms an important distinctive
criterion for hafting [18,20]. Other observations that should
have resulted in a correct interpretation of BT1 rest in the dis-
tinct presence of macroscopic scarring, which is patterned on
a microscopic level, the amount and distribution of bright
spots [17], the polish distribution (e.g. limited to outer ridge),
the absence of a well-developed polish on the dorsal right edge
(if hand-held, the fingers would have been located there and
would have caused a distinct polish formation), the presence
of a clear wood-like polish on the dorsal left fracture edge
(butt). The only remaining argument in favor of the proposed
interpretation is the unequal presence of traces on both lateral
edges, which is indeed typical for prehension [20], but which
is in this case a consequence of tool morphology.
6.1.2. Error 2: localization of haft limit too close to the
working edge
A correct localization of the haft limit appears to be prob-
lematic. In several cases it was located too close to the work-
ing edge. This is the result of basing interpretations too much
on dorsal ridge polish. The frequent placement of the finger
distal of the haft during use made this evidence unreliable.
The evidence on the edges proves to be more accurate. Regu-
larly, the limit was correctly identified on the edges, but the
ridge evidence was erroneously used to locate the maximal
extension of the haft towards the working edge.
6.1.3. Error 3: wood versus antler haft
Distinctions between a hafting polish resulting from antler/
bone or from wood proved to be difficult (e.g. BT6: antler in-
stead of wood; BT7: wood instead of antler, BT8: bone instead
of wood). Even upon re-examination, these polishes prove to
be hard to distinguish, despite the slightly rougher aspect of
antler polish in comparison to wood polish. It is clear that oth-
er wear traces need to be incorporated if progress is to be
made. A wood contact, for instance, typically results in stria-
tions parallel to the ridge and situated just next to it (Plate
10). Such striations were observed on BT6, but their value
as a distinctive criterion was not yet realized at the time.
Such criteria might be helpful in the future to prevent errors,
but confident distinctions between wood and antler seem haz-
ardous overall. Confusion is, however, understandable given
the very hard and dry wood used for hafting. It implies that
the kind of contact between the stone tool and the haft is com-
parable in both cases.
6.1.4. Error 4: use
The wrong (uncertain) interpretation of the use of BT1 can-
not be separated from the overall misinterpretation of the pre-
hensile mode. The initially inferred worked material was in
fact bone/antler and not schist, but based on the pronounced
rounding this interpretation was discarded. At the time, it
was not realized that bone/antler drilling could result in an in-
tense rounding and great importance was attached to its pres-
ence. The very hard nature of the antler worked may have
contributed to the formation of a rounding. A re-examination
7
of the use-wear evidence confirms that too much importance
was attached to the macroscopic analysis in which schist
was proposed as the most likely contact material based on
the presence of a clear gloss. The polish is actually a rather
typical bone/antler polish, especially on the dorsal ridge.
The interpretation of the action as grooving instead of drilling
was based on the limited use damage: only one scar could be
distinguished on the ventral tip. Apparently, its characteristics
were not examined in depth, as its initiation from the right side
of the tip is suggestive of drilling. The limited damage never-
theless remains bizarre, as most corresponding experimental
examples show important tip damage or even breakage. In ad-
dition, scars are generally laterally initiated and thus provide
indubitable evidence for the use motion. This observation
should caution one not to rely too much on the amount of
use damage. Scar characteristics are more decisive.
For the interpretation of the worked material of BT3, the
wear evidence was insufficiently conclusive to decide between
wood and ochred hide (Plate 11). The worked material in ques-
tion (leather) lies at the basis of this problem. Tanned leather is
generally not scraped, while hide in various stages is. It is, how-
ever, not impossible and it may perhaps be undertaken occa-
sionally to soften the leather. Additional experiments attest
that leather use-wear indeed resembles wood. No experimental
reference was available for this at the time and the analyst was
unaware of the close resemblance of leather and wood use-
wear. Only the presence of a clear rounding was suggestive
of hide working. A re-examination of the use-wear indicates
that the polish is somewhat rougher and more intrusive than
would be expected for wood. It further has a pitted instead of
a domed appearance and tiny grooves are present perpendicular
to the edge instead of gentle undulations. Maybe the scarring
evidence could have prevented the mistake since hide/leather
working results in less scarring in comparison to woodworking.
6.1.5. Error 5: haft contact
On BT8, the haft contact was interpreted as dorsal instead
of ventral. This is a minor mistake that was guided mainly by
a lateral fracture. In order for this fracture to form during
hafted use, the haft contact had to be dorsal on a narrow
haft. In reality, the fracture was produced during knapping.
While it was realized that the most proximal fracture was pos-
sibly produced during knapping, the more distal fracture was
not expected to be contemporaneous. This resulted in a biased
view on the hafting wear and an improper evaluation of count-
er-evidence.
6.1.6. Error 6: fixation method
The fixation method of BT 6 was interpreted as a leather
wrapping fixed with vegetal bindings instead of leather bind-
ings only. Given that both wear patterns do not really differ
on a microscopic level, this error can be dismissed as irrele-
vant. It was mainly guided by smoother traces in the medial
zone, which were actually due to wood. If the haft limit had
been identified more proximal, this mistake would not have
been made.
 8
Table 2
Blind test 2: experimental details and proposed interpretation of the wear traces for each analytical approach used ( included experimental data are correctly inferred based on wear analysis,  no data are
provided by the functional analyst, X irrelevant category for the tool in question)
Results BT9 BT10 BT11 BT12 BT13

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Tool type Endscraper Endscraper Tanged burin Endscraper Burin
Interpretation Interpretation Interpretation Interpretation Interpretation
Exp Exp Exp Exp Exp
Macrosco High Macrosc Low High Macrosc Low High Macrosc Low High Macrosc Low High
details Low power details details details details
pic power opic power power opic power power opic power power opic power power
proximal distal distal
Used part distal part + + + + + + + + + none + + + + + +

ARTICLE IN PRESS
part part part
U Worked fresh bone/ bone/ bone/
wood + + + wood + + + bone none + + + dry antler wood wood
s material bone antler antler antler
e Action adzing + + + adzing + + + grooving + + + none + + + grooving + + +
few short or short or short or moderate moderate moderate moderate moderate
Use duration 35min moderate 30min + + X + + + 55min moderate
seconds moderate moderate moderate or long or long or long or long or long

+
Hand-held /
hafted + + + hafted + + + hafted + + + none + + + hafted + + +

MODEL
hafted
Hafted / hand- proximal proximal proximal
+ + + distal part + + + + + + X + + + + + +
held part part (tang) part
Haft limit 3mm 29mm 30mm 46mm 36mm 37mm 39mm 24mm 24mm 24mm 24mm X + + + 20mm 20mm 20mm 24mm
P antler or antler or wood (or antler (or antler/ antler (or bone/ bone/
Haft material wood antler wood + antler X + + + antler +
r wood wood antler) bone) bone bone) antler antler
e all
Contact zone both or both
h both faces ventral ventral both faces + + (mainly + + + X + + + + + +
haft dorsal faces
e dorsal)
n Wrapping X leather + + X + + + X + + + X + + + X + + +
s Contact zone
X both faces + + X + + + X + + + X + + + X + + +
i wrapping
o Bindings leather + + X + + + X + + + X + + + leather
n Contact zone
none + + X + + + X + + + X + + + none
bindings
/ Fixation X + + + resin + + + X + + + X + + + resin + + +
Contact zone ventral
X + + + both faces + + X + + + X + + + all + + +
h fixation (or both)
a
f Haft type male (or
male split juxtaposed juxtaposed male split male male male male + + + X + + + male split male male male
t juxtaposed)
i Hafting
n method direct indirect + + indirect + + + direct + + + X + + + indirect + + +
g latero- latero-
Tool placement + + + + + + terminal + + + X + + + terminal + + +
distal distal
Tool direction transversal + + + transversal + + + axial + + + X + + + axial + + +
Orientation perpendi perpendi perpendi perpendi
+ + + + + + + + + X + + + + + +
active part cular cular cular cular
 Results BT14 BT15 BT16 BT17 BT18
Tool type Endscraper Endscraper Tanged endscraper Burin Burin
Interpretation Interpretation Interpretation Interpretation Interpretation
Exp Exp Exp Exp Exp
Macrosco High Macrosc Low High Macrosc Low High Macrosc Low High Macrosc Low High
details Low power details details details details
pic power opic power power opic power power opic power power opic power power
proximal distal distal
Used part distal part + + + distal part + + + + + + + + + + + +
part part part
Worked fresh hide+? bone/ fresh bone/ dry bone/ bone/ bone/
U earth + + + + wood bone wood antler wood? wood
material bone or wood antler bone antler antler antler antler antler
s
scraping/ grooving
e Action hoeing + + + scraping + + + scraping + + + grooving + + + planing planing
planing / planing
long or short or moderate
Use duration +30min + very long + 40min + 35min + + + 60min moderate moderate 60min + moderate +
very long moderate or long
Hand-held / hand-
hafted + + + hafted + + + hafted + + + + + hafted + + +
hafted held
Hafted / hand- proximal proximal distal proximal proximal
+ + + + + + + + + + + + + + +
held part part part part part part
P Haft limit

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

46mm 50mm 48mm 44mm 24mm 24mm 32mm 32mm 22mm 22mm 22mm 22mm X + + 29mm 35mm
r
wood or perhaps bone/ bone/ bone/ bone/ bone/ bone/
e Haft material wood + + antler wood antler X + + wood
antler wood antler antler antler antler antler antler
h
Contact zone perhaps both both
e both faces + + ventral ventral both both + + + X + both + + +
haft both faces faces
n

ARTICLE IN PRESS
Wrapping X + + + X + + + X + + + X + + + X + +
s
Contact zone
i X + + + X + + + X + + + X + + + X + +
wrapping
o
Bindings leather + + leather + + + X + + + X + + + X + +
n
Contact zone
edges + + dorsal + + + X + + + X + + + X + +

+
bindings
/

MODEL
Fixation X + resin + X + + + X + + resin X + resin + resin + +
Contact zone both
h X + both + X + + + X + + X + + all + +
fixation faces
a
f male or juxtapose male male
male split + juxtaposed not male male + + + X + + male + +
t Haft type male split d split split
i
n Hafting direct + indirect + direct + + + direct + + indirect X + + indirect + +
g method
Tool placement terminal + + + terminal + + + terminal + + + X + terminal + terminal + +
Tool direction axial + + + axial + + + axial + + + X + axial + axial + +
Orientation perpendic perpendi perpendi perpendi perpendi
+ + + + + + + + + X + + + +
active part ular cular cular cular cular

9
 ARTICLE IN PRESS
+ MODEL

10 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Table 3
Blind test 3: experimental details and proposed interpretation following an integrated functional analysis of the wear traces ( included experimental data are
correctly inferred based on wear analysis,  no data are provided by the functional analyst, X irrelevant category for the tool in question)
Results BT19 BT20 BT21 BT22 BT23 BT24
Tool type Tanged burin Blade Endscraper Tanged burin Endscraper Burin
Interpre Exp Interpretat Interpr Interpretat Interpretat Interpret
Exp details Exp details Exp details Exp details Exp details
tation details ion etation ion ion ation
Used part distal part + lateral + distal part + distal part + distal part + distal part +
Worked earth (incl. moderately
U material dry antler + dry antler bone/antler + fresh bone bone/antler fresh bone bone/antler +
herbs) dry wood
s grooving
e Action grooving + sawing + hoeing + grooving and scraping + grooving +
perforating
Use duration about 30min + 30min + 30min + 40min + 35min + 1h 15min +
Hand-held /
hafted + hafted + hafted + hafted + hafted + hafted +
hafted
P
Hafted / hand- proximal
r proximal part + + proximal part + proximal part + proximal part + proximal part +
held part part
e
h Haft limit 24mm 24mm 52mm 51mm 44mm 45mm 34mm 34mm 30mm 30mm 44mm 44mm
e Haft material antler + antler + wood + bone bone/antler wood + wood +
n Contact zone both
all + ventral both faces + ventral both all + dorsal +
s haft faces
i Wrapping leather X + X + X + X + leather +
o Contact zone
all X + X + X + X + ventral +
n wrapping
Bindings X + leather + leather + leather X leather leather +
/ Contact zone
X + edges dorsal edges + dorsal X edges edges +
bindings
h Fixation X + X + X + X + X + X +
a Contact zone
f X + X + X + X + X + X +
fixation
t Haft type male + male split juxtaposed male split + juxtaposed male male male split male split juxtaposed
i Hafting method indirect direct + direct + direct + direct + direct +
n Tool placement terminal + terminal + latero-distal + terminal + terminal + terminal +
g Tool direction axial + axial + axial + axial + axial + axial +
Orientation
perpendicular + parallel + perpendicular + perpendicular + perpendicular + perpendicular +
active part

To conclude, it is clear that the errors made in this test are
restricted to one main problem (schist polish) and two second-
ary problems (haft limit and haft material). It strongly suggests
that further improvement of test results is attainable. In this
stage, test results are sufficiently positive to justify an optimis-
tic attitude regarding the possibilities of inferring hafting and
the exact hafting arrangement on experimental and archaeo-
logical material. It is thus justified to argue that hafting traces
are interpretable. While the method as it was applied above
was still being developed and surely inadequate to limit or
even prevent errors, it allows a narrow focus of further experi-
ments on the highlighted problems. Whether or not these ad-
ditional experiments prove to prevent further errors can be
evaluated in the subsequent tests.
6.2. Blind test 2
The wear analysis of the second blind test is undertaken
at three separate, consecutive levels e macroscopic, low
power and high power e and an interpretation is proposed
for each approach used. Interpretations were not cross-
checked for potential contradictions and the information re-
trieved with each approach was separated as much as
possible. This procedure allows comparisons between the re-
sults obtained for each analytical level and an assessment of
the potential of each approach for identifying and interpret-
ing haft wear.
Experimental details and the wear interpretations obtained
through each analytical approach are included in Table 2. The
results are discussed per analytical level (Tables 5e7). In
some cases there is no interpretation provided for a certain
category (marked as -). This should not necessarily be
considered as a mistake. After all, an analytical approach
should not be judged based on the inability of providing
data it never pretended to be able to provide. For instance,
a macroscopic analysis generally does not allow for the inter-
pretation of the hafting material. When an interpretation is
provided that proves wrong, it is obviously counted as a
mistake.
The scores in Tables 5e7 depend on the importance of the
mistake. For instance, an inexact evaluation of the relative du-
ration is not considered of main importance for the final score
when the determination of other aspects of use (used tool part,
worked material, action) is correct. For the hafting arrange-
ment, the correct interpretation of the haft type and the haft
material are considered to be of main importance.
 ARTICLE IN PRESS
+ MODEL

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18 11

Table 4 some cases, while no macroscopic interpretation may have

Blind test 1: results of the wear analysis (0 wrong, 1 correct, been provided. Secondly, the certainty level of the interpreta-
0.5 partially correct interpretation; e irrelevant category for the tool in
question)
tions increased significantly, which is of key importance for
results to be reliable.
Results BT1 BT2 BT3 BT4 BT5 BT6 BT7 BT8 Total (/8)
Used part 1 1 1 1 1 1 1 1 8 6.2.3. High power level
Worked 0 1 0.5 1 1 1 1 1 6.5 The best results are undoubtedly obtained on a high power
material
Action 0 1 1 1 1 1 1 1 7
level. A complete interpretation for each tool is possible and
Relative 1 1 1 1 1 1 1 0 7 reliable hafting material identifications are provided. If used
duration in isolation, a low power analysis is only reliable for determin-
Use 0 1 0.5 1 1 1 1 1 6.5 ing the relative hardness of the hafting material, as is the case
for use-wear. A low power interpretation of the hafting ar-
Hafting 0 1 1 1 e 1 1 1 6
Prehension e e e e 0 e e e
rangement, however, is certainly equally reliable, if not more
reliable than a high power one.
Hafted/hand- 1 1 1 1 1 1 1 1 8
held part
6.2.4. Discussion
Haft limit 0 1 1 1 0 0 0 0 3
Haft material 0 1 1 1 0 0 0 0 3 The errors of the first blind test are used as a starting point
Contact 0 1 1 1 0 1 1 0 5 for the discussion of the results. It is examined whether or not
zone haft additional experimenting succeeded in resolving the interpre-
Wrapping 0 1 1 1 0 1 1 0 5 tative problem and whether new interpretative problems
Contact zone 0 1 1 1 0 1 1 0 5
were highlighted.
wrapping
Bindings 0 1 1 1 0 1 1 1 6
Contact zone 0 1 1 1 0 1 1 0 5 6.2.4.1. Error 1: prehension versus hafting. No problems re-
bindings main for the distinction between hand-held and hafted tools.
Fixation 0 1 1 1 0 1 1 1 6 Only for BT17 there was some disagreement in interpreta-
Contact zone 0 1 1 1 0 1 1 1 6
tion between the macroscopic and high power level on the
fixation
one hand and the low power level on the other hand. While
Haft type 0 1 1 1 0 1 1 1 6 the hand-held use was correctly inferred in the former cases,
Hafting method 0 1 1 1 0 1 1 0 5
Tool placement 0 1 1 1 0 1 1 1 6
the evidence was considered insufficiently conclusive in the
Tool direction 0 1 1 1 0 1 1 1 6 latter case. Hafting formed the most viable interpretation
Orientation AP 0 1 1 1 0 1 1 1 6 even though no particular hafting arrangement could be pro-
Interpretation 0 1 1 1 0 0.5 0.5 0.5 4.5 posed. The fact that only the haft material (i.e. bone/antler)
could be inferred is very significant as this interpretation is
based on polish mainly. A bone/antler polish, however, also
forms on the hand-held part of a bone/antler working tool
6.2.1. Macroscopic level resulting from the friction between the hand, which gets
The scores of the macroscopic analysis are surprisingly gradually covered with bone/antler particles that detach dur-
high, but these should be placed in their correct perspective. ing use, and the stone tool [20]. The fact that the error does
Above all, the results should be confronted with the certainty not concern the interpretation of the hafting arrangement is
level of each interpretation. Several interpretations were pro- very significant given that the results of the first blind test
posed with limited certainty (0e1) and should perhaps be con- already indicated that a low power analysis is less suitable
sidered as educated guesses. For the important categories such for reliably inferring the hafting material than it is to
interpretations are marked with grey shading in the tables. The reliably infer the hafting arrangement. It is evident that
high scores are also influenced by the experience gained the error is easily avoided when analytical methods are
throughout this research and the relatively formal tools under combined.
consideration. An important reason behind the potential suc-
cess of a macroscopic analysis of prehensile wear is the impor- 6.2.4.2. Error 2: localization of haft limit too far distal. The
tance of patterning that is frequently visible macroscopically. number of correct haft limit localizations significantly in-
It permits distinctions between hafted and hand-held tools creased in comparison with the previous blind test. While
and allows some propositions concerning the most likely haft- less than half of the haft limits were correctly assessed in
ing arrangement. test 1, the majority is now correctly identified, independent
of the analytical approach used. High power analysis proved
more accurate: a haft limit could be located on all hafted
6.2.2. Low power level tools and it proved to be correct for four out of eight cases.
At first sight, the low power scores are poorer than the mac- In only one case is the error more than 8 mm, so the results
roscopic ones. However, this view is biased. Firstly, the low appear to be relatively accurate. Experience may perhaps al-
power interpretation is more detailed, but proved wrong in low further improvements. Only on BT9 the haft limit was
 ARTICLE IN PRESS
+ MODEL

12 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Table 5
Blind test 2: results of the macroscopic analysis (0 incorrect; 0.5 partially incorrect or incomplete; 1 correct interpretation e irrelevant category for the
tool in question; shaded categories a degree of certainty was provided during the functional analysis; shaded cells uncertain interpretation)
Results BT9 BT10 BT11 BT12 BT13 BT14 BT15 BT16 BT17 BT18 Total (/10)
Used part 1 1 1 1 1 1 1 1 1 1 10
Worked material 1 1 1 1 0 1 0 0 1 0.5 6.5
Action 1 1 1 1 1 1 1 1 1 1 10
Relative duration 0.5 1 1 1 1 1 1 1 0.5 1 9
Use 1 1 1 1 0.5 1 0.5 0.5 1 0.5 8
Hafting 1 1 1 x 1 1 1 1 e 1 10
Prehension e e e x e e e e 1 e
Hafted / hand-held part 1 1 1 1 1 1 1 1 1 1 10
Haft limit e 0 1 1 1 1 1 1 1 e 7
Haft material 0.5 e 1 1 1 1 0 0 1 e 5.5
Contact zone haft 0.5 e 1 1 1 1 0 1 1 e 6.5
Wrapping 0 1 1 1 1 1 1 1 1 e 8
Contact zone wrapping 0 1 1 1 1 1 1 1 1 e 8
Bindings 0 1 1 1 1 1 1 1 1 e 8
Contact zone bindings 0 1 1 1 1 1 1 1 1 e 8
Fixation 1 1 1 1 1 1 1 1 1 e 9
Contact zone fixation 1 1 1 1 1 1 1 1 1 e 9
Haft type 0.5 1 1 1 0.5 1 0.5 1 1 e 7.5
Hafting method 0 1 1 1 1 1 1 1 1 e 8
Tool placement 1 1 1 1 1 1 1 1 1 e 9
Tool direction 1 1 1 1 1 1 1 1 1 e 9
Orientation AP 1 1 1 1 1 1 1 1 1 e 9
Interpretation 0 0.5 1 1 1 1 0.5 0.5 1 0.5 7

truly erroneously located. A fracture during use had re-
moved most of the hafted portion and the broken piece
was not handed over to the analyst. Mistakenly, this possi-
bility was not really considered during the analysis, which
logically resulted in a wrong location of the haft limit.
Based on these results, it is valid to argue that haft limits
can be adequately identified and that the evidence visible on
the edges is indeed more reliable than the dorsal ridge
evidence.
6.2.4.3. Error 3: wood versus antler haft. The haft material
identifications improved, but the distinction between wood
and antler hafts remains problematic. Actually, only wood-
hafted tools cause problems, as all antler hafts are correctly
inferred (BT11, BT13, BT15). The animal nature of the
haft was even interpretable on a macroscopic level, except
for BT15. By contrast, wooden hafts seem to result in confus-
ing wear traces. The confusion is probably due to the hard
and dry nature of the wood that is used for hafting, which re-
sults in wear traces that more closely resemble antler wear
(Plates 12 and 13). This is exemplified by BT18 for which
no interpretation could be proposed on either a macroscopic
or low power level. On a high power level, an attempt was
made, but the indicated certainty level remained extremely
low. Antler haft wear more closely corresponds to what is ex-
pected for use-wear, which facilitates interpretations (Plate
14). Further experiments should thus focus on a better under-
standing of wood haft wear in comparison to wood use-wear
and not merely on the distinction between wood and antler haft
wear. The fact that wood haft wear can be sufficiently
distinctive is confirmed by its correct interpretation on BT10
and BT14 (Plate 15).
6.2.4.4. Error 4: use. Most tool use interpretations were cor-
rect, independent of the analytical approach used. All identifi-
cations of the used edge and the action were accurate. Only the
worked material determination caused some interpretative
problems and occasionally the relative use duration was not
correctly assessed, which is of lesser importance. Given that
only a high power analysis claims to allow confident material
identifications, those results are primarily discussed. Only the
worked material of BT17 proved problematic. A very limited
amount of use polish hampered a confident interpretation and
the indicated certainty level is thus very low. Antler was cor-
rectly inferred on a macroscopic and a low power level, but the
intrusion of some polish into the concavities was used as an
argument for hardwood on a high power level. In hindsight,
the analyst should have realized that wood causes a far more
intrusive polish and more importantly, that the scarring evi-
dence was more important for the material identification given
the poorly developed polish. The latter is demonstrated by the
low power results. In all other cases, high power analysis
proved more reliable than low power analysis for material
identifications, including accurate distinctions between wood
and antler working. For the other worked materials, no inter-
pretative problems arose.
6.2.4.5. Error 5: haft contact. No problems persisted for the
identification of the face that was in contact with the haft.
Errors only concerned distinctions between a haft contact
 ARTICLE IN PRESS
+ MODEL

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18 13

Table 6
Blind test 2: results of the low power analysis (0 incorrect; 0.5 partially incorrect; 1 correct interpretation; e irrelevant category for the tool in question;
shaded categories a degree of certainty was provided during the functional analysis; shaded cells uncertain interpretation)
Results BT9 BT10 BT11 BT12 BT13 BT14 BT15 BT16 BT17 BT18 Total (/10)
Used part 1 1 1 1 1 1 1 1 1 1 10
Worked material 1 1 1 1 0 1 1 1 1 0 8
Action 1 1 1 1 1 1 1 1 1 1 10
Relative duration 0.5 1 1 1 0.5 0.5 0 1 0 0.5 6
Use 1 1 1 1 0.5 1 1 1 1 0.5 9
Hafting 1 1 1 x 1 1 1 1 e 1 9
Prehension e e e x e e e e 0 e
Hafted / hand-held part 1 1 1 1 1 1 1 1 1 1 10
Haft limit 0 0 1 1 1 1 0 1 e e 5
Haft material 0.5 1 1 1 1 0.5 1 0 0 e 6
Contact zone haft 0 1 1 1 1 1 0 1 0 1 7
Wrapping 1 1 1 1 1 1 1 1 1 1 10
Contact zone wrapping 1 1 1 1 1 1 1 1 1 1 10
Bindings 1 1 1 1 1 0 1 1 1 1 9
Contract zone bindings 1 1 1 1 1 0 1 1 1 1 9
Fixation 1 1 1 1 1 0 1 1 0 1 8
Contact zone fixation 1 1 1 1 1 0 1 1 e 1 8
Haft type 0 1 1 1 0.5 1 0 1 e 1 6.5
Hafting method 1 1 1 1 1 0 1 1 e 1 8
Tool placement 1 1 1 1 1 1 1 1 0 1 9
Tool direction 1 1 1 1 1 1 1 1 0 1 9
Orientation AP 1 1 1 1 1 1 1 1 0 1 9
Interpretation 0 1 1 1 1 0.5 0.5 0.5 0 1 6.5

with both faces or with only one face (BT9, BT14 (high power),
BT15). Such determinations are linked with the interpretation
of the haft type and are dealt with below. Suffice it to state that
tool morphology (i.e. longitudinal convexity, transversal cross-
section, etc.) plays a significant role in this, as it determines
the amount of contact that is possible between the stone tool
and its haft and thus the wear formation.
6.2.4.6. Error 6: haft type. In contrast to test 1, not all haft
types were correctly inferred, but it needs to be stipulated
that not all mistakes are equally fundamental. A distinction be-
tween a male haft, in which the tool is mounted into a hole,
and a male split haft, in which the tool is mounted into a cleft,
is not always possible, in particular when resin is used, and
such mistakes are considered of secondary importance. Conse-
quently, the haft type interpretation of BT10 is marked as cor-
rect and also the error of BT13 is of limited importance. In the
latter case, the bindings made contact with the edges, which is
potentially identifiable based on the very particular scarring
that is caused by bindings [18,20]. However, due to the use
of resin in between the stone tool and the bindings, these scars
could not form and the use of bindings was thus nearly impos-
sible to identify.
The haft type of BT14 was only wrongly inferred using
high power microscopy. Two observations initiated this inter-
pretation: the clear difference in polish intensity between the
dorsal and ventral face, and the rougher and more intrusive na-
ture of the dorsal polish. The former is in fact a result of the
presence of cortex and the important longitudinal curvature
of the tool that hinders intense haft contact with both faces.
The latter is a result of intruding earth particles that dominated
the polish formation. The available evidence can thus account
for the mistake. For future interpretations, it is important to re-
alize the extent at which intrusive earth particles can influence
the hafting trace formation process. For BT15, a male split
haft (with restricted dorsal haft contact) was inferred instead
of a juxtaposed one. Both haft types can be difficult to distin-
guish if the dorsal haft contact is limited. The poor polish for-
mation on the dorsal face indicates that this was the case here.
In addition, the impact of bindings on the tools edges is com-
parable in juxtaposed and male split arrangements (Plate 16),
providing no valid arguments for a reliable distinction.
Consequently, in cases of poor trace development, distinc-
tions between juxtaposed and male split haft types will remain
difficult given their comparable impact on wear trace forma-
tion. This should be regarded as one of the interpretative limits
of the method. When traces are sufficiently developed, reliable
distinctions should be possible.
6.2.4.7. Error 7: fixation method. The fixation method was
overall correctly inferred. The few errors have been dealt
with indirectly above and are only summarized here. The
lack of interpreting bindings for BT13 was a consequence of
the use of resin, which made such an inference near to impos-
sible. For BT16 and BT17, the use of resin was incorrectly in-
ferred by a single analytical approach, but this had no impact
on the remainder of the interpretation. The limited amount of
visible wear, which is typical in the case of resin use, formed
the main argument. The occurrence of a somewhat rough sur-
face polish affected the interpretation of BT16, but this mis-
take can be avoided by reference to diagnostic friction spots
that are typical of resin use.
 ARTICLE IN PRESS
+ MODEL

14 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18

Table 7
Blind test 2: results of the high power analysis (0 incorrect; 0.5 partially incorrect; 1 correct interpretation; e irrelevant category for the tool in question;
shaded categories a degree of certainty was provided during the functional analysis; shaded cells uncertain interpretation)

Results BT9 BT10 BT11 BT12 BT13 BT14 BT15 BT16 BT17 BT18 Total (/10)
Used part 1 1 1 1 1 1 1 1 1 1 10
Worked material 1 1 1 1 1 1 1 1 0 1 9
Action 1 1 1 1 1 1 1 1 1 1 10
Relative duration 0.5 1 1 1 1 1 1 1 1 1 9.5
Use 1 1 1 1 1 1 1 1 0.5 1 9.5

Hafting 1 1 1 x 1 1 1 1 1
10
Prehension x 1

Hafted / hand-held part 1 1 1 1 1 1 1 1 1 1 10

Haft limit 0 0 1 1 1 1 0 1 1 0 6
Haft material 0 1 1 1 1 1 1 0 1 0 7
Contact zone haft 0 1 1 1 1 0.5 0 1 1 1 7.5
Wrapping 1 1 1 1 1 1 1 1 1 1 10
Contact zone wrapping 1 1 1 1 1 1 1 1 1 1 10
Bindings 1 1 1 1 1 1 1 1 1 1 10
Contact zone bindings 1 1 1 1 1 1 1 1 1 1 10
Fixation 1 1 1 1 1 1 1 0 1 1 9
Contact zone fixation 1 1 1 1 1 1 1 0 1 1 9

Haft type 0 1 1 1 0.5 0 0 1 1 1 6.5

Hafting method 1 1 1 1 1 1 1 0 1 1 9
Tool placement 1 1 1 1 1 1 1 1 1 1 10
Tool direction 1 1 1 1 1 1 1 1 1 1 10
Orientation AP 1 1 1 1 1 1 1 1 1 1 10

Interpretation 0.5 1 1 1 1 0.5 0.5 0.5 1 0.5 7.5

In conclusion, even though mistakes were made, the quality
of the results significantly improved in comparison to the first
blind test. Important is the assessment that distinctions be-
tween hand-held and hafted tools are reliable, which obviously
forms the most fundamental distinction with regard to prehis-
toric assemblages. Also on a more detailed level, confident in-
terpretations are possible for the hafting material and
arrangement. Interpretations proved to be most difficult for
wood-hafted tools.
6.3.1. Discussion
6.3.1.1. Error 1: prehension versus hafting. The scores of the
final test confirm that reliable distinctions between hand-held
and hafted tools are possible. No errors were made.
6.3.1.2. Error 2: localization of haft limit too far distal. The
haft limit was correctly inferred in all cases.

6.3.1.3. Error 3: wood versus antler haft. The haft material

6.3. Blind test 3
In the final blind test, all approaches are combined in order
to arrive at one interpretation per tool. The deficiencies of one
particular approach, as identified above, are to be compensated
by the advantages of another approach in order to minimize
errors. This procedure is presented as the most fruitful one
for adequately examining and interpreting prehensile wear
(and all other wear for that matter) on stone tools.
Experimental details and wear interpretations are included
in Table 3.
was correctly inferred in all cases, despite the use of antler
and in particular wooden hafts. The better understanding of
wood haft wear that could be obtained through additional ex-
periments proved to be sufficiently adequate.
6.3.1.4. Error 4: use. Tool uses were correctly inferred
in all cases.
6.3.1.5. Error 5: haft contact. Similar to the second test, the
few errors concern the distinction between male and
 ARTICLE IN PRESS
+ MODEL

V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18 15

Table 8
Blind test 3: results of the wear analysis (integrated approach) (0 incorrect;
0.5 partially incorrect; 1 correct interpretation; e irrelevant category for
the tool in question)
Results BT19 BT20 BT21 BT22 BT23 BT24 Total (/6)
Used part 1 1 1 1 1 1 6
Worked material 1 1 1 1 1 1 6
Action 1 1 1 1 1 1 6
Relative duration 1 1 1 1 1 1 6
Use 1 1 1 1 1 1 6
Hafting 1 1 1 1 1 1 6
Prehension e e e e e e
Hafted/hand-held 1 1 1 1 1 1 6
part
Haft limit 1 1 1 1 1 1 6
Haft material 1 1 1 1 1 1 6
Contact zone haft 1 0 1 0 1 1 4
Wrapping 1 1 1 1 1 1 6
Contact zone 1 1 1 1 1 1 6
Plate 10. Hafting striation parallel and just next to the ridge of BT6, associated
wrapping
with wood haft polish, mistakenly interpreted as antler polish (200
).
Bindings 1 1 1 0 0 1 5
Contact zone 1 0 1 0 0 1 3
bindings
Fixation 1 1 1 1 1 1 6
the tools edges and the haft, a factor that allows distinctions
Contact zone 1 1 1 1 1 1 6 between male and male split hafts. Given the lack of wear at-
fixation tributable to this e generally very intense e contact, the haft
Haft type 1 0 1 0 0.5 1 3.5 was interpreted as male split. This mistake is difficult to avoid,
Hafting method 1 1 1 1 1 1 6 but the lack of scars typical of the use of bindings should have
Tool placement 1 1 1 1 1 1 6 been considered more fully. However, these scars do not form
Tool direction 1 1 1 1 1 1 6 easily on retouched edges, in particular not when retouch is as
Orientation AP 1 1 1 1 1 1 6
coarse as is the case here.
Interpretation 1 0.5 1 0.5 0.5 1 4.5 To conclude, the test results are satisfactory. Few methods
can pretend to be without errors and some degree of error
should be accepted, in particular when these concern mainly
juxtaposed arrangements and thus depend on the amount of
secondary features. The distinction between hand-held and
contact between stone tool and haft. The mistake made for
hafted tools is definitely of primary importance and proved
BT20 is difficult to avoid given that the wear pattern for jux-
to be reliable. Even more detailed aspects proved to be inter-
taposed arrangements with bindings and the one for male split
pretable with a high degree of confidence. Only the haft type
arrangements with bindings are near to identical when wear
regularly caused some confusion. Given that hafting research
trace formation is poor, as is the case here.

6.3.1.6. Error 6: haft type. The main problem concerns the

identification of the haft type. As stated, the mistake is difficult
to avoid for tools like BT20 and BT24, due to the nearly iden-
tical impact on the tool of both hafting arrangements. Only
when wear traces are sufficiently developed a confident dis-
tinction can be made (cf. test 2).
For BT22 and BT23 the situation is different, even though
both mistakes are understandable. On BT22, scars typical of
the use of bindings were observed, but based on the inferred
use e perforating e these scars were interpreted differently.
There is one exceptional case in which scars typical of binding
use may form without the intervention of bindings, being
a male-hafted tool used as a perforator [16]. It is clear that the
use-wear evidence should have been examined more closely,
which would have made apparent that the grooving motion
was more important than the perforating motion, the latter
only being an occasional part of the use cycle.
In the case of BT23, the hole in the haft was actually too Plate 11. Use-wear polish from scraping tanned leather on the ventral scraper-
wide for the tool used. This prevented intense contact between head of BT3 (100
).
 ARTICLE IN PRESS
+ MODEL
16 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
Plate 12. Hafting wood polish on the butt of BT9, mistakenly interpreted as
antler polish (100
).
is only in its initial stages, improvements are to be expected
and a future decrease in the number of errors is still plausible.
The results of this test most certainly demonstrate that the
combination of a macroscopic, low power and high power ap-
proach for the analysis of stone tools and the interpretation of
prehensile (and other) wear is fruitful and allows reliable
inferences.
7. Conclusion
It is generally accepted that blind tests are very useful in mi-
croscopic investigations. They are part of a continuous process
towards better or more adequate interpretations and analysts
should test their interpretations at a regular basis independent
of their experience. In spite of that, a blind test centered on
Plate 13. Hafting wood polish on the ventral butt of BT9, mistakenly inter-
preted as antler polish (200
).
Plate 14. Hafting antler polish on the bulb of BT11 (200
).
prehensile wear was never attempted before. The presented se-
quence of blind tests has demonstrated that it is possible to dis-
tinguish between hand-held and hafted stone tools and that
even the hafting arrangement could be inferred correctly in
most cases. In addition the blind tests have shown the value
of an integrated approach with macroscopic and microscopic
low power and high power observations. Indicative macroscopic
hafting wear may form, but interpretations that are only based
on a macroscopic analysis proved unreliable and can only be
used as a starting point of a more profound investigation. While
a low power analysis proved to be very successful on the level of
the interpretation of the hafting arrangement, a high power
analysis is more successful in determining the contact material.
The latter corresponds to what was also observed for use-wear.
For the distinction between hand-held and hafted tools, both
low power and high power analysis proved successful even
Plate 15. Hafting wood polish on the ventral butt of BT10 (200
).
 ARTICLE IN PRESS
+ MODEL
V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
Plate 16. Sliced scars from the contact with leather bindings on the left ventral
medial edge of BT15 (25
).
though a combined analysis remains the most reliable approach
at all times given that the deficiencies of one method are com-
pensated by the other.
Evidently, these results need to be elaborated upon in the
future if improvements are to be made. For now, the experi-
mental reference collection that lies at the basis of this re-
search proved sufficient for deriving valid wear
characteristics to distinguish between prehension and hafting
and different hafting arrangements, but it is clear that the col-
lection will need to be elaborated continuously in order to re-
main reliable.
These blind test results have important implications for the
investigation of stone tool assemblages. While the presented
data are purely experimental in nature, they justify a successful
large-scale application of the integrated approach to archaeo-
logical assemblages [21,22]. When combined with use-wear
data, the investigation of prehensile wear allows a better and
more comprehensive insight in a tools life cycle [18]. Archae-
ological identifications of hand-held and hafted tools have im-
portant implications for the interpretation of tool design, site
function, origins of hafting and modern human behavior
[30], and the organization of technology.
To conclude, hafting research allows totally new insights,
unobtainable by other means with significant impact on ar-
chaeological interpretations. It involves a new approach that
can prove very successful, in particular when integrated with
use-wear research and other non-microscopic approaches
(e.g. technological and spatial analyses).
Acknowledgements
We are indebted to the Research Fund of the Katholieke
Universiteit Leuven for their financial support of this research
(OT97/25, OT01/10). We thank the Prehistosite de Ramioul
(Lie`ge, Belgium) and the Chercheurs de la Wallonie for the
17
provided facilities for performing our experiments. We thank
the reviewers for their valuable comments.
References
[1] P.C. Anderson-Gerfaud, Contribution methodologique a` lanalyse des
micro-traces dutilisation sur les outils prehistoriques, The`se de 3e cy-
cle, n. 1607, Institut du Quaternaire, Universite de Bordeaux I, Talence,
1981.
[2] D.B. Bamforth, G.R. Burns, C. Woodman, Ambiguous use traces and
blind test results: new data, Journal of Archaeological Science 17 (1990)
413e430.
[3] R. Fullagar (Ed.), A Closer Look: Recent Australian Studies of Stone
Tools, Archaeological Methods Series, vol. 6, Sydney University, Syd-
ney, 1998.
[4] P.A. Gendel, L. Pirnay, Microwear analysis of experimental stone tools:
further test results, in: D. Cahen (Ed.), Tailler! pour quoi faire: Prehis-
toire et technologie lithique II. Recent Progress in Microwear Studies,
Studia Praehistorica Belgica 2 (1982), pp. 251e265 (Tervuren).
[5] J. Kamminga, Over the Edge: Functional Analysis of Australian Stone
Tools, University of Queensland Anthropology Museum, Occasional Pa-
pers in Anthropology 12, 1982.
[6] L.H. Keeley, Experimental Determination of Stone Tool Uses: A Micro-
wear Analysis, University of Chicago Press, Chicago and London, 1980.
[7] L.H. Keeley, Hafting and retooling: effects on the archaeological record,
American Antiquity 47 (1982) 798e809.
[8] L.H. Keeley, M.H. Newcomer, Microwear analysis of experimental flint
tools: a test case, Journal of Archaeological Science 4 (1977) 29e62.
[9] K. Knutsson, R. Hope, The application of acetate peels in lithic usewear
analysis, Archaeometry 26 (1984) 49e61.
[10] T.H. Loy, Prehistoric blood residues: detection on tool surfaces and iden-
tification of species of origin, Science 220 (1983) 1269e1271.
[11] M.H. Newcomer, R. Grace, R. Unger-Hamilton, Investigating microwear
polishes with blind tests, Journal of Archaeological Science 13 (1986)
203e217.
[12] G.H. Odell, Toward a more behavioral approach to archaeological lithic
concentrations, American Antiquity 45 (1980) 404e431.
[13] G.H. Odell, The application of micro-wear analysis to the lithic component of
an entire prehistoric settlement: methods, problems, and functional recon-
structions, Unpublished PhD thesis, Harvard University, Cambridge (MA),
1977.
[14] G.H. Odell, The mechanics of use-breakage of stone tools: some testable
hypotheses, Journal of Field Archaeology 8 (1981) 197e209.
[15] G.H. Odell, F. Odell-Vereecken, Verifying the reliability of lithic use
wear assessment by Blind Test: the low power approach, Journal of
Field Archaeology 7 (1980) 87e120.
[16] V. Rots, Hafting traces on flint tools: possibilities and limitations of
macro- and microscopic approaches, Unpublished PhD thesis, Katholieke
Universiteit Leuven, 2002a.
[17] V. Rots, Bright spots and the question of hafting, Anthropologica et Prae-
historica 113 (2002b) 61e71.
[18] V. Rots, Towards and understanding of hafting: the macro- and micro-
scopic evidence, Antiquity 77 (2003a) 805e815.
[19] V. Rots, The identification of experimental microscopic hafting traces on
lithic tools, in: Le Secretariat du Congre`s (Ed.), Sessions Generales et
Posters/General Sessions and Posters, Acts of the XIVth UISPP Congress
University of Lie`ge, Belgium, 2e8 September 2001 (2003b), pp. 47e55
BAR International Series 1145.
[20] V. Rots, Prehensile wear on flint tools, Lithic Technology 29 (2004)
7e32.
[21] V. Rots, Wear traces and the interpretation of stone tools, Journal of Field
Archaeology 30 (1) (2005) 61e73.
[22] V. Rots, P. Van Peer, Early evidence of complexity in lithic economy:
core-axe production, hafting and use at Late Middle Pleistocene site
8-B-ll, Sai Island (Sudan), Journal of Archaeological Science, in press.
doi:10.1016/j.jas.2005.08.002.
 ARTICLE IN PRESS
+ MODEL
18 V. Rots et al. / Journal of Archaeological Science xx (2006) 1e18
[23] V. Rots, P.M. Vermeersch, Experimental characterisation of microscopic haft-
ing traces and its application to archaeological lithic assemblages, in:
E.A. Walker, F. Wenban-Smith, F. Healy (Eds.), Lithics in Action. [28]
Papers from the Conference Lithic Studies in the Year 2000, Lithic Studies
Society Occasional Paper 8, Oxbow Books, Oxford, 2004, pp. 156e168.
[24] V. Rots, L. Pirnay, P. Pirson, O. Baudoux, P.M. Vermeersch, Experimental [29]
Hafting Traces. Identification and Characteristics, Notae Praehistoricae
21 (2001) 129e137.
[25] S.A. Semenov, Prehistoric Technology Translated by M.W. Thompson, [30]
Cory, Adams and Mackay, London, 1964.
[26] J.J. Shea, On accuracy and relevance in lithic use-wear analysis, Lithic
Technology 16 (1987) 44e50. [31]
[27] J.J. Shea, Methodological considerations affecting the choice of analyti-
cal techniques in lithic use-wear analysis: tests, results and application,
in: S. Beyries (Ed.), Industries Lithiques, Traceologie et technologie,
vol. 2 (1988), pp. 65e82 BAR International 411.
R. Tringham, G. Cooper, G.H. Odell, B. Voytek, A. Whitman, Experi-
mentation in the formation of edge-damage: a new approach to lithic
analysis, Journal of Field Archaeology 1 (1974) 171e196.
G. Unrath, L.R. Owen, A. Van Gijn, E.H. Moss, H. Plisson, P. Vaughan,
An evaluation of use-wear studies: a multi-analyst approach, Early Man
News 9/10/11 (1986) 117e175.
P. Van Peer, V. Rots, J. Vroomans, A story of colourful diggers and grind-
ers. The Sangoan and Lupemban at site 8-B-11, Sai Island, Northern
Sudan, Before Farming (online version) (2004) article 1.
L. Wadley, M. Lombard, B. Williamson, The first residue analysis blind
tests: results and lessons learnt, Journal of Archaeological Science 31
(11) (2004) 1491e1501.